Abstract: Abstract Objective The purpose of the present retrospective single centre study is to assess the possible causes of these infections. Patients and methods Two hundred and twenty-eight patients were included in the period from January 2003 until April 2011. Patient and disease demographics,as well as EVD data, and the occurrence of infection were reviewed, compared, and included in a risk-analysis study.Results The population's mean age was 56士15 years and the sexes were equally distributed. Most frequently,the indication for EVD was hydrocephalus due to in-traventricular haemorrhage (48.2%).An infection was documented in 23.2% of all patients.Duration of EVD drainage appeared to be a risk factor for infection(>11 days: OR 4.1;95% CI 1.8一9.2, p=0.001).CSF sampling frequency was also a significant risk-factor (no sampling: OR 0.2, 95% CI 0.2-0.5, p=0.003).Conclusions We found a relatively high percentage of EVD-related infections.After multivariate analysis there appears to be a relation with duration of drainage，frequent CSF sampling and cerebrospinal fluid leakage.As a result, a new EVD protocol is proposed in our institution that we believe will decrease the number of EVD-related infections to a minimum.

Key words: External ventricular drainage, central nervous system infection, protocol.